Get Started
Home
Topics
Search
Library
Research questionHow can we measure whether transformer representations distinguish senses of the same word across contexts?The same written word can express different senses across domains, making it difficult to determine whether contextual representations separate meanings rather than merely reflect word identity or broad topic differences. A meaningful test must hold the word form fixed while varying its labeled context.
AI
Evaluation & Benchmarks
Mechanistic Interpretability
Natural Language Processing
Latest papersRecent research connected to this question, newest first.Technical Manual for a Toolkit for Measuring Contextual Individuation in Transformer Language ModelsThe source documents an open toolkit using labeled bridge forms, Wikipedia corpora, occurrence localization, layer-wise representation extraction, domain-pairwise separation measurements, and paired visualizations. It is a methodological and implementation reference and reports no empirical outcomes for particular transformer models or bridge-form sets.research paper · Sep 4, 2026
Related questions
How can we compare feature contributions to language-model representations when those features are correlated?How can multilingual representation sharing be measured without confusing anisotropy with genuine cross-lingual structure?How can we evaluate whether text-to-image models express visual metaphors across domains?Does reusing Transformer layers improve language-model quality when parameter, compute, and KV-cache budgets are matched?